|
|
|
Question : sorting top output
|
|
Hi Experts
I have earlier posted a question regarding sorting of top output. The response i got almostgave me the output i was looking for . Im posting the question and response again so that someone of you can follow it up and suggest me an apt solution..
On my server there is an overnight batch job that runs and sometimes it crashes in the night. Using a scrip that was already in place, i collected the reports on overnight batch job which will run "Top" command every 5 minutes throughout the night. But the output is completely jumbled and i dont have time to run the top command again tonight to get the output properly. So i was hoping to sort out the data back into the format of TOP command.. I am not worried abt getting the exact format, but will be good enough to have the data like this...
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
3676 user1 8 0 37952 62m 1920 S 0.0 6.1 0:00.12 bash
3640 user1 8 0 31296 49m 768 R 0.0 4.8 0:00.01 top
Its almost impossible for me to manually sort the data, so i was wondering if i can use some script or command that will sort of the data below in the above format.
The below is a sample from the Top statistics recorded over night.
#####################################################################################
End process stats: 26/06/2007 22:00:01
Start process stats: 26/06/2007 22:00:01
load averages: 0.10, 0.14, 0.21 22:05:01 208 processes: 205 sleeping, 1 zombie, 2 on cpu Memory: 4096M real, 126M free, 1675M swap in use, 8627M swap free PID USERNAME THR PRI NICE SIZE RES STATE TIME CPU COMMAND 22344 administrator 4 52 2 78M 76M sleep 1:20 0.00% iengine 21969 administrator 4 52 278M 76M sleep 1:21 0.00% peng 22438 administrator 56 29 2 49M 29M sleep 0:42 0.01% java 3193 administrator 10 52 2 24M 20M sleep 0:06 0.00% CAdapter_ 3189 administrator 9 53 2 23M 19M sleep 0:04 0.00% CAdapter_ 3279 administrator 13 53 2 20M 18M sleep 0:06 0.00% sdk2etx 19871 administrator 13 0 6 44M 16M sleep 719:26 0.00% rvd 21277 administrator 11 52 2 16M 14M sleep 0:13 0.00% app1 21215 administrator 11 52 2 16M 14M sleep 0:13 0.00% app1 21255 administrator 11 52 2 16M 14M sleep 0:12 0.00% app1 21284 administrator 10 53 2 16M 14M sleep 0:13 0.00% app1 21339 dministrator 11 52 2 16M 14M sleep 0:13 0.00% app1 21193 administrator 10 53 2 16M 14M sleep 0:13 0.02% app1 21361 administrator 11 52 2 16M 14M sleep 0:13 0.00% app1 21378 administrator
10 53 2 16M 14M sleep 0:13 0.00% app1 21316 administrator 11 52 2 16M 14M sleep 0:12 0.00% app1 21234 administrator 11 53 2 16M 14M sleep 0:12 0.00% app1 3240 administrator 12 52 2 15M 13M sleep 0:02 0.00% app1 437 root 12 58 0 18M 12M sleep 31:52 0.00% python 21694 administrator 4 53 2 19M 12M sleep 7:33 0.00% peng 8951 administrator 10 58 0 18M 12M sleep 1:15 0.00% tix_node 17347 administrator 18 0 0 38M 11M sleep 2:13 0.00% java 9069 administrator 4 52 0 14M 11M sleep 1:41 0.00% etx_logsrv 5176 administrator 10 52 2 19M 11M sleep 28:40 0.04% tix_node 21751 oracle 1 52 0 18M 10M sleep 0:00 0.00% oracle 17188 administrator 1 52 0 18M 10M sleep 1:22 0.00% oracle 17211 administrator 5 58 0 15M 9936K sleep 1:58 0.00% etx_gds_ora 18575 administrator 8 58 031M 9856K sleep 46:30 0.00% tix_node 18424 administrator 8 58 0 21M 9752K sleep 60:06 0.00% tix_node 17230 administrator 5 59 0 14M 9336K sleep 0:51 0.00%
End process stats: 26/06/2007 22:05:01
Start process stats: 26/06/2007 22:05:01
####################################################################################
Ozo gave me the following solution which sorts out all the output in the correct format,
#perl -lne 'print for /\d+\s+\w+\s+(?:\d+\s+){3}(?:\d+\w\s+){2}\w+\s+[\d:]+\s+[\d.]+%\s+\w+/g'
But the only problem is that i have the output without the timestamps which is again making it hard to analyze the data:
3354 admin 4 58 0 8840K 6896K sleep 0:18 0.00% app1
17475 admin 4 58 0 8768K 6824K sleep 0:05 0.00% app1
593 root 7 58 0 10M 6816K sleep 1:32 0.00% jre
9068 admin 5 58 0 10M 6656K sleep 0:01 0.00% ogsrv
17219 admin 4 58 0 14M 6640K sleep 1:44 0.00% gds_ora
17221 admin 5 59 0 13M 6568K sleep 1:15 0.00% gds_ora
7018 admin 5 52 2 7664K 6520K sleep 0:01 0.00% wtix
27507 admin 4 52 2 7656K 6512K sleep 0:00 0.00% wtix
16560 oracle 11 59 0 19M 6472K sleep 0:03 0.00% oracle
16562 oracle 11 59 0 19M 6424K sleep 0:33 0.00% oracle
####################################################################################
Can someone / ozo please suggest how i can also grep the time and date with the output. Thanks
I would really appreciate any response towards this as this is extremely important for me.
Thanks
|
Answer : sorting top output
|
|
perl -lne 'print for /(\d+\s+\w+\s+(?:\d+\s+){3}(?:\d+\w\s+){2}\w+\s+[\d:]+\s+[\d.]+%\s+\w+|^(?:Start|End) process stats:.*)/g'
|
|
|
|
|
|
|
